Item Selection
Item selection is a key design decision because it determines which questionnaire items become nodes in the PCN.
Design goal
The algorithm should balance:
- clinical relevance,
- person specificity,
- comparability across participants,
- and assessment burden.
A working target is approximately 3–6 nodes, with an upper limit that keeps the number of pairwise causal judgments manageable.
Candidate selection strategies
Severity-driven selection
Select the highest-rated questionnaire items above a predefined threshold.
Advantages:
- simple,
- reproducible,
- easy to automate.
Limitation:
- assumes severity equals personal relevance.
Participant-driven selection
Present elevated/highest-rated items and ask which are currently most relevant or important.
Advantages:
- stronger person specificity,
- may better reflect subjective priorities.
Limitation:
- introduces an additional choice process.
Hybrid selection
A promising default:
- identify candidate items based on questionnaire severity;
- show the candidate set to the participant;
- ask which problems are currently relevant;
- retain up to a prespecified maximum.
This preserves a strong connection to the validated questionnaire while allowing participant relevance to influence the final network.
Methodological comparison
Alternative selection procedures can themselves be validated.
Possible comparisons:
- severity-only vs hybrid selection,
- top-4 vs top-6 nodes,
- fixed threshold vs rank-based selection.
Outcomes:
- burden,
- representativeness,
- reliability,
- incremental validity.
Open decisions
See Decisions for choices that remain to be finalized.